摘要
在充分考虑人耳听觉特性和噪声统计特性的基础上,提出一种时频结合Bark尺度自适应阈值的语音消噪算法,在Bark频域上自适应调整增强系数可以较准确地进行阈值判定。仿真实验验证,时频结合算法在低信噪比输入情况下较传统语音降噪方法具有明显优势,其在消除高斯白噪声的同时有效降低了语音损失,可获得最大信噪比,谱失真测度最小,增强语音的MOS(Mean Opinion Score)评分明显提高,具有较好的听觉效果。
Based on fully considering the human auditory characteristics and the noise statistical characteristics, a time-frequency combination Bark scale adaptive threshold algorithm for speech de-noising is presented. In the Bark frequency domain the adaptive adjustment of coefficient can increase the accuracy of the threshold value. Simulation results show that the time-frequency combination algorithm in low SNR input cases has obvious advantages over the traditional algorithm, such as to achieve maximal signal-to-noise ratio and minimizing spectral distortion, by using the new algorithm, the damage of weak speech signal can be avoided effectively and the noise eliminated adequately at the same time. The Mean Opinion Score (MOS) of the enhanced speech has a better performance in Subjective test and a better auditory effect could be obtained.
出处
《应用声学》
CSCD
北大核心
2010年第4期256-262,共7页
Journal of Applied Acoustics
关键词
小波包消噪
自适应阈值算法
时频结合
Bark尺度
Wavelet packet de-noising, Adaptive threshold algorithm, Time frequency combination, Bark scaled